Dr W. Schwizer, Department of Gastroenterology and Hepatology, University Hospital of Zürich, Rämistrasse 100, CH-8091 Zürich, Switzerland. E-mail: email@example.com
Background : The intermittent loss of oil or liquid faeces (‘spotting’) is an adverse effect that occurs in obese patients during treatment with the lipase inhibitor orlistat; the pathophysiology is unknown.
Aim : To investigate the effects of orlistat on anorectal sensorimotor function and continence.
Methods : Obese subjects susceptible to spotting were identified by an unblind trial of orlistat. Obese spotters (n = 15) and non-spotters (n = 16) completed a randomized, double-blind, cross-over trial of orlistat and placebo. Anorectal function was assessed by rectal barostat and anal manometry, together with a novel stool substitute retention test, a quantitative measurement of faecal continence.
Results : Orlistat increased stool volume and raised faecal fat and water. Treatment had no effect on anorectal motor function, but rectal sensation was reduced; on retention testing, the volume retained was increased. Subjects susceptible to spotting had lower rectal compliance, heightened rectal sensitivity and weaker resting sphincter pressure than non-spotters. On retention testing, gross continence was maintained; however, spotters lost small volumes of rectal contents during rectal filling.
Conclusion : Treatment with orlistat has no direct adverse effects on anorectal function or continence. Spotting occurs during treatment with orlistat when patients with sub-clinical anorectal dysfunction are exposed to increased stool volume and altered stool composition.
Orlistat is an effective therapeutic option for weight reduction and maintenance in overweight and obese patients.1,2 Taken with meals, orlistat inhibits gastric and pancreatic lipase and reduces fat digestion by up to 30%.3 In combination with dietary restriction, this effect facilitates weight loss, has favourable effects on a variety of risk factors for cardiovascular disease and slows the development of type 2 diabetes.1,2,4 This treatment is generally well tolerated, with a safety profile similar to placebo, with the exception of gastrointestinal adverse effects. One of these is ‘spotting’, the loss of small amounts of oil or liquid faeces from the anorectum. Spotting occurs in approximately one-fifth of patients taking 120 mg orlistat with meals, and requires the withdrawal of medication in 3.6% of patients.2 The pathophysiology of spotting is unknown.
In order to discover the cause of continence problems, an understanding of the normal mechanisms governing faecal continence is required. Weak anal sphincter function is the most common cause of gross faecal incontinence presenting to specialist clinics; however, spotting occurs in patients with no history of incontinence,1,2 and it is well known that sphincter pressures are normal in a significant minority of incontinent patients.5–7 Other factors that may be of importance include the capacity of the faecal reservoir,5,8,9 anorectal function including rectal compliance,7–10 anorectal sensitivity7,9–11 and the physical properties of stools.5,12 The lack of an objective, quantitative measurement of faecal continence has limited previous attempts to assess the relative importance of these factors in health and disease. To address this issue, we combined comprehensive physiological testing of anal sphincter and rectal function with a novel retention test developed and validated in our department. This technique delivers a viscous stool substitute into the rectum by continuous infusion and provides a quantitative assessment of anorectal sensation and continence during rectal filling. Preliminary evidence has shown that stool substitute retention test measurements of anorectal sensitivity and continence are sensitive to normal variations in rectal capacity and anal sphincter function.13,14
We hypothesized that anorectal sensorimotor function is affected by treatment with orlistat, and that such effects are the cause of continence problems in this population. Healthy, obese subjects were studied in a randomized, double-blind, cross-over trial of orlistat against placebo. This design allowed a separate assessment to be made of the effects of orlistat on the anorectum and the intrinsic features of anorectal function that may predispose to spotting when the continence mechanism is challenged.
Materials and methods
Obese (body mass index > 30 kg/m2) subjects were recruited by advertisement. Elderly patients (> 55 years) and women who had given birth were excluded to avoid possible confounding factors such as occult degeneration or obstetric damage of the sphincter. Subjects with a history of continence problems, gastrointestinal disease, surgery (except appendectomy), psychological problems or abnormalities on laboratory testing were also excluded. No subject on regular medication (except contraception) was included. A negative pregnancy test was required from all women of childbearing potential. All participants signed an informed consent to participate in these studies, which were approved by the ethical committee of the University Hospital of Zürich.
Prior to randomization, obese subjects underwent an unblind assessment of continence when treated with orlistat. Orlistat 120 mg (F. Hoffmann La Roche Ltd., Basel, Switzerland) was administered three times per day with standard meals for 2 days. Administration of the morning dose mixed in 40 g of butter increased the likelihood of adverse events related to lipase inhibition. Capsules were provided for lunch and dinner with standard meals to ensure consistent results. Subjects recorded adverse events and were classified as spotters if any unintentional loss of oil or stool was observed over a 3-day period.
Within 2 weeks after the spotting assessment, subjects were randomized and entered a double-blind, placebo-controlled, cross-over study. The hospital pharmacy dispensed medication according to a computer-generated randomization list. Subjects received orlistat 120 mg or placebo three times a day with meals for 4 weeks. All doses were self-administered. The treatment periods were separated by a 2-week washout period. Subjects were instructed to keep their usual diet during the study. Gastrointestinal adverse events were recorded in a daily diary during treatment periods. Prior to the anorectal studies, obese subjects recorded food intake (6 days) and performed a stool collection (4 days).
A sample size of 16 spotters and 16 non-spotters was determined prospectively to allow the detection of a difference of 10 mmHg in sphincter pressure at a level of alpha = 0.05 and beta = 0.1, assuming a standard deviation of 20 mmHg.15 Fifty-three obese subjects were screened: 10 were excluded on the basis of a body mass index of less than 30 kg/m2 (n = 4), co-morbid conditions (n = 3), lack of contraception (n = 1) and withdrawal of consent (n = 2). Thus, 43 entered the spotting assessment and were classified as spotters (n = 19) and non-spotters (n = 24). Four non-spotters were not randomized because 16 subjects had already completed the study. There were no differences in baseline characteristics between non-spotters and spotters in terms of the body mass index [36.6 kg/m2 (30.3–49.9 kg/m2) vs. 36.1 kg/m2 (30.8–42.2 kg/m2)], age [43 years (29–54 years) vs. 39 years (20–53 years)] or sex (male to female ratio: 8 : 8 vs.6 : 9). Eight subjects (four from each group) failed to complete the study. One (spotter) withdrew following continence problems on orlistat, and the remainder withdrew due to coincident accident or illness (n = 3), non-compliance with the protocol (n = 3) and family problems (n = 1). One spotter withdrew just before the planned study completion and was not replaced. Reference values for the investigations were established in 10 normal weight subjects (male to female ratio, 5 : 5; age, 20–35 years; body mass index, 19–25 kg/m2).
At the end of each treatment period, subjects underwent anorectal studies. Subjects arrived for the anorectal studies following a minimum 4-h fast. Clinical observations and digital rectal examination were performed. If stool was noted, the subject was instructed to evacuate. Bowel preparation was not required in any patient.
The barostat technique was applied in accordance with standard recommendations.16 A balloon with infinite compliance up to a maximum volume of 800 mL (length, 10 cm) was connected to an electronic barostat (Distender series II, G&J Electronics Inc., Toronto, Canada). Following assessment of the minimal distending pressure and a conditioning expansion, the pressure–volume relationship was measured at 2-mmHg increments in the range of 0–40 mmHg. Rectal sensations of urgency, wind, discomfort (rectal pressure) and pain were assessed by 100-mm visual analogue scale (VAS) measurements during the second minute of phasic balloon distension (30 mL/s) at 12, 24, 36 and 42 mmHg in random order.
Anal sphincter manometry was acquired using a four-channel, low-compliance water perfusion technique in accordance with standard recommendations.17 Resting pressure was measured during a pull-through with a radial catheter. A minimum of three attempts of voluntary anal contraction over 40 s was recorded with a longitudinal catheter with channels at 1-cm intervals through the sphincter.
The retention test was performed by rectal infusion of a viscous suspension with the consistency of soft stool (viscosity at 37 °C, 10 000 cP). This substance (Stocki, Knorr, Switzerland) is the standard preparation at our institution for magnetic resonance defecography. Flexible tubing with a rectal introducer was placed such that the tip was 8 cm proximal to the anal verge. The subject was seated on a commode with a graduated cylinder placed under the seat. A peristaltic pump (Watson-Marlow 505S) transferred the stool substitute from a container to the subject and a digital scale monitored the volume delivered. Rectal infusion proceeded at 60 mL/min to a maximum of 1500 mL. The subject was instructed to retain the enema for as long as possible. During rectal filling, the anorectal sensation, volume delivered and volume lost were recorded until losses exceeded the capacity of the cylinder (250 mL).
Stool was analysed for volume, wet and dry weight, total fat and free fatty acids. Underwear liners which contained faecal matter were weighed. Orlistat was measured in the stool to document adherence to the study protocol.
The self-reported incidence of gastrointestinal events was assessed from the subject diary, including stool frequency and consistency (Bristol classification) and adverse gastrointestinal symptoms of urgency, spotting and gross faecal incontinence.
The pressure–volume relationship obtained by the barostat technique (Figure 1) provides an indirect assessment of wall stiffness and resistance to distension during rectal filling. At lower pressures, the balloon volume rises rapidly as the rectum expands into the potential lumen and the muscle wall relaxes to ‘accommodate’ the balloon. Rectal compliance was defined as the maximum slope of the pressure–volume curve.16 At higher pressure, further increases in volume are slower, occurring against the passive resistance of the rectal wall. Asymptotic compliance was defined as the slope of the linear continuation. Rectal capacity was defined as the intra-balloon volume at 40 mmHg; this novel measurement is highly reproducible with low variance and is not affected by sequence, rate or type of barostat distension.13
The VAS score of rectal sensations increased with intra-balloon pressure in an approximately linear relation for all sensations. The increase in VAS score per unit pressure was used for pairwise comparisons. For the inclusion of rectal sensation in the analysis of variance (anova), the VAS scores were offset-log transformed to stabilize variance and an aggregate score was calculated.
The resting pressure of the anal sphincter was defined as the maximum pressure recorded during the pull-through, averaged over the four manometry channels. During the squeeze attempt, the pressure decreases as a result of external sphincter fatigue. The maximum squeeze pressure was defined as the highest recorded channel pressure during a 40-s attempt of voluntary anal contraction relative to the resting pressure. A robust linear regression was fitted to these pressure data. The ‘initial squeeze’ pressure was defined by extrapolating the regression line to the start of the squeeze attempt. The ‘fatigue rate’ was defined as the gradient of a robust regression line fitted to these pressure data. The squeeze duration was defined as the time in seconds for fatigue from the initial squeeze pressure to the baseline resting pressure. Maximum and initial squeeze pressures were strongly correlated (r = 0.87); the initial squeeze pressure was preferred for statistical analysis because it is derived from the entire data set and has lower variance.
The primary end-points recorded during the retention test were the volume infused at first sensation and urgency, the volume at first incontinence (Vi) and the maximum volume retained (Vm). As shown in the representative data presented in Figure 2, Vm approximates a stable reservoir volume in the majority of subjects. This volume is closely correlated with the rectal capacity as assessed by the barostat (see validation below). Thus, in continent subjects, Vm appears to be determined in large part by the structural limitations of the rectum and can be taken as an assessment of gross continence function. However, the large variation in rectal capacity (range, 200–600 mL) makes it difficult to compare faecal continence between individuals. The difference between the maximum volume retained and the volume at first incontinence (Vm − Vi) is less sensitive to this confounding factor and can be used to express continence during rectal filling, i.e. the ability to retain the enema up to the reservoir volume without volume loss. A low value for this parameter indicates that continence is well maintained during rectal filling.
Spotters were compared with non-spotters using a mixed model anova, with ‘subject’ as a random variable and ‘treatment group’ and ‘spotter status’ as fixed variables, for between- and within-group analysis. Data were considered to be significant at alpha < 0.05. Normal weight subjects provided reference values only.
Demographic data and quantitative variables are presented as medians (interquartile range). The results presented in the tables are group median averages calculated by anova and differ slightly from the raw data presented in the figures.
Table 1 presents a comparison of spotters and non-spotters. Table 2 summarizes the effects of orlistat on anorectal function and continence in obese subjects.
Table 1. Anorectal function and continence in spotters and non-spotters. Data show median with interquartile range. Reference values from normal weight subjects
VAS, visual analogue scale.
* Values for continence during rectal filling (Vm − Vi) were highly skewed and thus the median values do not equal the difference between the median values for Vm and Vi.
The minimal distending pressure was comparable in both groups of obese patients. The rectal compliance was lower in spotters than in non-spotters (P < 0.05) and the asymptotic compliance and rectal capacity were similar in both groups (Figure 3a). During treatment with orlistat, there was a non-significant increase in rectal compliance in spotters and non-spotters; no consistent effects were seen for other parameters. Spotters recorded a higher VAS score for each sensation than non-spotters (Figure 3b); the aggregate VAS score confirmed the higher rectal sensitivity (P < 0.01). Orlistat reduced the VAS score of wind and discomfort (Figure 3b); the aggregate VAS score demonstrated lower rectal sensitivity during treatment (P < 0.05).
The resting anal pressure was higher for non-spotters than spotters (P < 0.05). The maximal anal squeeze pressure (P = 0.09) and initial squeeze pressure (P = 0.08) tended to be higher in non-spotters. The fatigue rate and squeeze duration were similar in both groups. Orlistat had no effect on sphincter function. No difference was found in the length or symmetry of the anal canal for either comparison (data not shown).
First awareness was sensed at lower volumes by spotters than non-spotters (P < 0.04); no difference was seen in the urge to defecate. Orlistat decreased the anorectal sensitivity during rectal filling. Subjects reported the first perception at higher infusion volumes during active treatment (P < 0.04); similar findings were seen for urgency (P = 0.08).
Representative examples of the retention test measurements of continence from non-spotters and spotters are displayed in Figure 2 (nine of 10 normal weight subjects behaved like non-spotters). At volumes approximating the maximum volume retained (Vm), a stable ‘reservoir volume’ was demonstrated in 84% of all subjects (42 of 50 completed tests with at least three episodes of volume loss). Vm was similar in non-spotters and spotters; however, the volume at first incontinence (Vi) was lower in spotters. This finding was significant (Vm − Vi, P < 0.03) when the variation in rectal capacity/reservoir volume was taken into account (Figures 4 and 5). Impaired continence during rectal filling (Vm − Vi) was observed on at least one test day in 12 of 15 spotters, but in only two of 16 non-spotters (P < 0.01 for χ2 comparison). On orlistat, Vm and Vi increased (P < 0.01), an effect due mainly to an increase in retained volumes for non-spotters (Figure 4). Continence during rectal filling was not affected by treatment.
Validation of retention test data
The reproducibility of retention test measurements was demonstrated in eight normal weight subjects investigated on two occasions separated by approximately 4 months. The coefficient of variation for first sensation was 0.67, for urgency 0.77, for Vi 0.73 and for Vm 0.63. The parameter Vm − Vi (continence during rectal filling) was < 100 mL on both test days for seven of the eight subjects. Vm correlated strongly with rectal capacity as assessed by the barostat (r2 = 0.95, P < 0.001). This relationship was also significant in obese subjects (r2 = 0.65, P < 0.01). Anorectal sensation and continence during rectal filling were also sensitive to anal sphincter function.14
The recovery of orlistat from the stool demonstrated satisfactory adherence to the study protocol in both groups (data not shown). The weight of stool increased during treatment with orlistat [placebo, 173 g (125–211 g); orlistat, 246 g (163–333 g); P < 0.002]. Total fat (P < 0.001) and water (P < 0.01) content were markedly increased on treatment, but the fat-free dry weight remained unchanged.
Spotters had somewhat higher stool weight in both treatment phases [spotter, 204 g (170–230 g); non-spotter, 186 g (95–261 g)], with a relatively higher proportion of fat and water in the stool than in the non-spotter group. Covariant analysis revealed an association between the water content of the stool and Vm − Vi (continence during rectal filling). Including the factor ‘water content of the stool’ in the anova, the effect of the factor ‘spotter’ on Vm − Vi (continence during rectal filling) was more pronounced with the correction (P < 0.003) than without (P < 0.02).
During the 4-week treatment phases, subjects reported spotting or incontinence more frequently on orlistat than on placebo [3 (0–9) vs. 0 (0–1); P < 0.01]; overall adverse events were also increased on active treatment [8 (1–26) vs. 0 (0–1); P < 0.001]. One subject (non-spotter) reported a high rate of faecal loss during both treatment phases (an extreme outlier on Grubb's test). Orlistat was not measured in the placebo phase stool and questioning revealed major problems with diary completion; the patient was excluded from further analysis of the study diary.
During active treatment, spotters reported higher rates of spotting and incontinence than non-spotters [8 (2–17) vs. 2 (0–3); P < 0.03]; overall adverse events were also increased [16 (8.5–30) vs. 2.5 (2–17); P < 0.02]. Both passive (insensible faecal loss) and urge incontinence were reported by the subjects.
This study did not confirm the hypothesis that anorectal sensorimotor function is adversely affected by treatment with orlistat. Instead, the findings demonstrated sub-clinical anorectal dysfunction in obese subjects susceptible to spotting.
Treatment with orlistat had no significant effects on anorectal motor function; however, anorectal sensation was decreased during phasic barostat distensions and rectal filling. This effect did not impair continence; on the contrary, active treatment appeared to facilitate rectal filling as assessed by the stool substitute retention test. Although reduced anal and rectal sensitivity has often been reported in incontinent patients,9,11,18 increased rectal sensitivity has also been associated with continence problems, including frequent defecation and urge incontinence.7,8,10 Moreover, a reduction in rectal sensitivity has been linked to improved continence function in experimental and clinical settings.6,19 Thus, orlistat did not impair anorectal motor function, and faecal spotting is very unlikely to be caused by the reduction in rectal sensitivity observed. Instead, the enhanced ability to retain the stool substitute during orlistat treatment suggests that the anorectum adapts to the ‘challenge’ of the increased stool volume, fat and water which occurs during active treatment.
Obese subjects susceptible to spotting had evidence of anorectal dysfunction. The resting anal pressure was lower in this group and there was a trend to a weaker anal squeeze pressure. Low resting pressure is characteristic of patients with passive incontinence.20 Low squeeze pressure is typical in patients with urge incontinence.20 Subjects reported both forms of faecal loss during orlistat treatment. The mixed pattern of sphincter weakness and faecal loss documented in spotters probably reflects the important interaction of internal and external anal sphincter function in continence function.20,21
The rectal compliance was lower in spotters than in non-spotters. Compliance is an indirect assessment of the resistance (stiffness) of the rectum to distension. A lower compliance indicates that the rectal pressure increases more rapidly early during rectal filling, and suggests that the active relaxation of the rectum to a volume load (rectal accommodation) may be impaired in spotters. Other explanations are less likely because the passive properties of the rectum, as assessed by the asymptotic compliance and rectal capacity (Figure 1), were comparable in both groups. In addition to the lower compliance, a heightened rectal sensitivity was noted in spotters. This combination implies increased reactivity of the rectum.8 Previous authors have reported similar findings in association with faecal incontinence;7,10,22 however, anatomical defects of the anal sphincter were common in these patients and the changes may have been secondary to abnormal reservoir function.22 This study demonstrates that abnormal rectal compliance and sensitivity (reactivity) may affect continence in subjects with anal sphincter function within the normal range.
The stool substitute retention test quantified the impact of anorectal dysfunction on faecal continence. The maximum volume retained (Vm) was closely associated with the rectal capacity as assessed by the barostat. Moreover, at volumes close to Vm, the ongoing volume loss approximated to the ongoing volume delivered, and the retained volume remained constant (Figure 3). Together, these findings indicate the presence of a stable faecal reservoir volume in continent subjects, the magnitude of which is determined by the anatomical limitations of the rectum (rectal capacity). Gross continence can be considered as the ability to retain the stool substitute up to the anatomical limit (the reservoir volume ≈ Vm). Continence during rectal filling (Vm − Vi) can be considered as the ability to retain the enema up to the reservoir volume without volume loss. The latter may be determined by functional rather than structural factors, including rectal compliance and anal sphincter function. Both spotters and non-spotters retained the stool substitute up to a stable reservoir volume; in contrast, the majority of spotters lost small volumes of stool substitute as the retention test progressed (Figures 2 and 5). These findings suggest that gross continence function was preserved in spotters, but continence during rectal filling was impaired. In addition, during treatment with orlistat, the increase in the volume retained on retention testing was less evident for spotters than non-spotters (Figure 4). Thus, the anorectal dysfunction in spotters may be compounded by a relative failure of the continence mechanism to adapt to the ‘challenge’ of the increased stool volume and altered stool composition caused by lipase inhibition.
Although measurements of anorectal physiology and continence demonstrated significant differences between spotters and non-spotters, spotters did not show gross abnormalities comparable with those in patients presenting with clinical incontinence. Spotting occurred only during treatment with orlistat, and it was only then that spotters reported more adverse events than non-spotters. Thus, the anorectal dysfunction in spotters was sub-clinical, becoming apparent only when the continence mechanism was ‘challenged’ during active treatment. This is consistent with the observation that stool consistency affects continence; many clinical patients are incontinent to loose stool, but retain control of solids.5,12 Although spotters reported more faecal loss than non-spotters, the severity of spotting on orlistat was highly variable. The imperfect classification of the subjects by the screening test may partly explain this observation; however, alterations in stool properties may also be important. Increasing the dose of orlistat or the amount of fat in the diet both have graded effects on the composition and consistency of the stool and increase the risk of spotting.3 Standard diets were imposed for the spotting assessment, but could not be imposed during the 4-week cross-over study. Wide variations of fat intake were recorded. A formal assessment was not performed; however, non-spotters appeared to have an appreciable risk of spotting on a high-fat diet, whereas spotters had a relatively low risk of spotting on a low-fat diet. Further evidence of an interaction between continence and stool properties was provided by the association between high stool water content (i.e. low consistency) and impaired continence during rectal filling (Vm − Vi).
This study identified the pathophysiology of spotting by combining comprehensive physiological measurements with objective assessment of continence during rectal filling. The stool substitute retention test allowed us to quantify the effects of the physiological differences between spotters and non-spotters on continence function. In contrast with previous retention tests using saline,5 viscous fluids9 or solids,23 this investigation accurately identified subjects with continence problems. Moreover, retention test measurements correlated with rectal capacity and anal sphincter function,13,14 variables known to be important in continence function. These advantages may originate from the continuous (rather than bolus) infusion of a viscous stool substitute, a technique that stimulates the anal canal and rectum in a comparatively physiological fashion. We believe that this investigation shows promise as an objective measurement of continence function in health and disease.
In conclusion, spotting occurs when subjects with sub-clinical anorectal dysfunction are exposed to the effects of orlistat on the stool. Susceptible individuals have lower anal sphincter pressure, heightened rectal reactivity and may adapt less effectively to the challenge of increased stool volume, fat and water to the continence mechanism. Orlistat treatment itself has no direct adverse effects on anorectal physiology. These findings suggest that measures to reduce anorectal reactivity, increase internal anal sphincter pressure and regulate stool consistency may reduce the severity of continence problems on orlistat. The clinical relevance of this research may extend beyond the management of faecal spotting. Impaired rectal compliance and abnormal rectal sensitivity and reactivity have all been reported in association with faecal incontinence.7–10 Hitherto, the relevance of these findings has been unclear because most patients have also shown gross abnormalities of the anal sphincter. This study confirms that such abnormalities are sufficient to impair faecal continence, especially when present in combination and in the presence of loose stools.
This study was supported by a grant from the Swiss National Science Foundation and F. Hoffmann La Roche Ltd. We would like to thank Dr Hans Lengsfeld for his advice on study design, stool analysis and evaluation, and Professor Jim Brasseur for his valuable insights into the biomechanics of rectal function.